JPH07160271A - Adaptive active silencer - Google Patents

Abstract

PURPOSE:To provide an adaptive active silencer which always prevents howling and does not degrade a silence performance even if sound field characteristics between a loudspeaker for silence and a sensor microphone and/or between the loudspeaker for silence and a monitor microphone are different at the time of identification and at the time of silence. CONSTITUTION:In an adaptive active silencer in which a noise is reduced by making a sound wave having same amplitude as that of a noise detected by a sensor microphone 1 and an opposite phase with a filter 5 for silence and superimposing it to a noise, further characteristics of the filter 5 for silence are successively updated by using a synthetic sound consisting of a sound projected by a loudspeaker for silence detected by a monitor microphone 9 and a noise, this device is provided with a means which successively updates characteristics of a filter for identifying a sound field between a loudspeaker 8 for silence used for preventing howling and a sensor microphone, a means which successively updates characteristics of a filter for identifying a sound field between a loudspeaker 8 for silence used for preventing howling and a monitor microphone, and a generating means of a signal which is made an input to two filter characteristics updating means for identifying a sound field.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an active silencer for reducing noise by adding artificially created sound to noise.

[0002]

2. Description of the Related Art Sound created based on detected noise is
BACKGROUND ART Active acoustic control, in which a sound field is controlled by artificially adding the noise, is widely used for noise reduction, so-called muffling. It is called active noise reduction in which noise is reduced by superimposing a sound wave having the same amplitude and opposite phase as the noise. In active muffling, if the amplitude and phase characteristics of the sound field to be muffled change during muffling, it is necessary to make the characteristics of the muffling device follow the change, and adaptive control theory is used for this purpose. Conventionally, an active muffling device using the adaptive control theory has been performed, and it is disclosed in, for example, Japanese Utility Model Laid-Open No. 3-70490 as a device applied to muffling a duct system of an air conditioner.

The main part of this device is shown in FIG. In FIG. 5, 0 is a duct, 1 is a sensor microphone for detecting noise propagating from the left in the duct 0, 2 is an amplifier, 3 is an analog-digital (AD) converter, 4 is a subtractor, 5 is usually Filter for noise reduction realized by Finite Impulse Response (FIR) filter, 6 digital-analog (DA) converter, 7 amplifier, 8 noise reduction speaker, 9
Is a monitor microphone for detecting a combined sound of radiation sound and noise from the muffling speaker 8, 10 is an amplifier, 11 is an AD converter,
Reference numeral 12 is a filter coefficient updating unit of the muffling filter 5, 13 is an identification filter for describing the characteristic of the sound field from the muffling speaker 8 to the sensor microphone 1 which is usually realized by the same FIR filter as the muffling filter 5, Reference numeral 14 is an identification filter for describing the characteristics of the sound field from the noise elimination speaker 8 to the monitor microphone 9, which is usually realized by the same FIR filter as the noise elimination filter 5 and the identification filter 13, and 15 is a control device.

The noise propagates in the duct 0 from left to right, is detected by the sensor microphone 1 and amplified by the amplifier 2 on the way, and is then converted into a digital signal by the AD converter 3 in the controller 15. . The converted digital signal is input to the + side input terminal of the subtractor 4. The output signal of the subtractor 4 is converted into a signal having the same amplitude as the noise and an opposite phase by passing through the noise elimination filter 5, is converted back to an analog signal by the DA converter 6, and is amplified by the amplifier 7.
The sound-deadening speaker 8 radiates the sound into the duct 0 as sound.

A synthesized sound of the sound emitted from the noise elimination speaker 8 and the noise is detected by the monitor microphone 9, amplified by the amplifier 10, and converted into a digital signal by the AD converter 11. The converted digital signal is input to the filter coefficient updating unit 12 as an error signal, and the filter coefficient of the silencing filter 5 is updated so that this error always approaches the minimum value. Least Mean Squar as coefficient update algorithm
Well-known adaptive algorithms such as the e (LMS) algorithm are used.

Since the radiated sound of the muffling speaker 8 is also included in the detection signal of the sensor microphone 1, the muffling speaker 8
The sound field characteristic between the sensor microphone 1 and the sensor microphone 1 in advance, that is, the identification filter 13 is obtained by identifying the sound field characteristic.
Input signal to the minus side input terminal of the subtractor 4 and subtraction from the detection signal of the sensor microphone 1 to obtain the input signal of the noise elimination filter 5 for the emitted sound of the noise elimination speaker 8. To prevent the howling from occurring.

As input data to the adaptive algorithm in the filter coefficient updating unit 12, the sound field characteristic between the muffling speaker 8 and the monitor microphone 9 is subtracted to the identification filter 14 obtained by measuring or identifying in advance. The sound field noise between the muffling speaker 8 and the monitor microphone 9 is eliminated at the time of muffling by using the signal passed through the output signal of the device 4.

[0008]

In the above-mentioned conventional apparatus, the temperature in the duct at the time of identification is different from that at the time of muffling,
If the wind speed in the duct changes during the muffling as in the case of the air conditioning duct, the muffling speaker 8 and the sensor microphone 1 and / or the muffling speaker 8 and the monitor microphone 9 at the time of identification and at the time of muffling. The sound field characteristics will be different. If the sound field characteristics between the muffling speaker 8 and the sensor microphone 1 are different between the identification and the muffling, the subtraction by the subtractor 4 cannot completely prevent the radiated sound of the muffling speaker 8, resulting in howling. Will occur. Further, if the sound field characteristics between the muffling speaker 8 and the monitor microphone 9 are different, the difference in characteristics is mixed in the filter coefficient updating unit 12 as sound field noise, and the muffling performance is deteriorated. An object of the present invention is to prevent howling at all times even when the sound field characteristics between the muffling speaker and the sensor microphone and / or between the muffling speaker and the monitor microphone differ between identification and muffling, and do not reduce the muffling performance. An object is to provide an adaptive active silencer.

[0009]

According to the present invention, a sound wave having the same amplitude and opposite phase as that of noise detected by a sensor microphone is created by a silencing filter and superposed on the noise to reduce the noise. Sound field identification between a muffling speaker and sensor microphone used to prevent howling in an adaptive active muffling device that sequentially updates the characteristics of the muffling filter by using the detected sound of the muffling speaker as the error signal Means for sequentially updating the characteristics of the filter for sound, a means for sequentially updating the characteristics of the filter for identifying the sound field between the muffling speaker and the monitor microphone used for removing the sound field noise, and the above 2 for performing the sound field identification. And a signal generating unit that is an input to one filter characteristic updating unit.

[0010]

When the muffling is being executed, a signal such as minute noise is superimposed on the output of the muffling filter and is output to the muffling speaker. Using this minute signal radiated from the noise reduction speaker, the sound field between the noise reduction speaker and the sensor microphone and the sound field between the noise reduction speaker and the monitor microphone are sequentially identified by the adaptive algorithm. Representing the sound field characteristics between the muffling speaker and the sensor microphone and the sound field characteristic between the muffling speaker and the monitor microphone, which were identified before execution of the muffling 2
By sequentially replacing the two identification filters with the respective identification filters obtained by the above identification, it is possible to prevent howling at all times even if the characteristics of the sound field change due to temperature changes etc. after the start of muffling. The sound deadening performance can be maintained.

The present invention will now be described in detail with reference to the drawings. FIG. 1 is a block diagram showing the outline of an embodiment of the present invention. In FIG. 1, 1 is a sensor microphone for detecting noise propagating from the left, 2 is an amplifier, 3 is an analog-digital (AD) converter, 4 is a subtractor, and 5 is usually a Finite Imp.
Filter for noise reduction realized by ulse response (FIR) filter, 6 is a digital-analog (DA) converter, 7 is an amplifier, 8 is a speaker for noise reduction, 9 is a combined sound of radiation sound and noise from the noise reduction speaker 8. Monitor microphone for detecting the noise, 10 is an amplifier, 11 is an AD converter, 12 is a filter coefficient updating unit of the noise reduction filter 5, and 13 is a sensor from the noise reduction speaker 8 which is usually realized by the same FIR filter as the noise reduction filter 5. An identification filter that describes the characteristics of the sound field reaching the microphone 1, 14 is a sound field reaching the monitor microphone 9 from the sound deadening speaker 8 that is usually realized by the same FIR filter as the sound deadening filter 5 and the identification filter 13. For identifying the characteristics of the above, 15 for a control device, 16 for a signal generator, 17 for an adder, 18 for sound field identification between the muffling speaker 8 and the sensor microphone 1. Filter, the filter coefficient update unit of the identifying filter 18 is 19, 20 is a subtracter, sound field identification filter between mute speaker 8 and the monitor microphone 9 21, 22 the filter coefficient update unit of the identifying filter 21,
23 is a subtractor.

In the present embodiment, the noise propagating from a noise source (not shown) is detected by the sensor microphone 1 and is amplified by the amplifier 2.
After being amplified by, the signal is converted into a digital signal by the AD converter 3 in the control device 15. The converted digital signal is input to the + side input terminal of the subtractor 4. The output signal of the subtractor 4 is converted into a signal having the same amplitude as the noise and an opposite phase by passing through the silence filter 5, and is input to one input terminal of the adder 17. The output signal of the adder 17 is D
The analog signal is returned to the analog signal by the A converter 6 and the amplifier 7
After being amplified by, the sound is emitted as sound from the muffling speaker 8. The combined sound of the radiated sound and noise from the muffling speaker 8 is detected by the monitor microphone 9, amplified by the amplifier 10, and converted into a digital signal by the AD converter 11. The converted digital signal is input to the filter coefficient updating unit 12 as an error signal, and the filter coefficient of the silencing filter 5 is updated so that this error always approaches the minimum value. Least Mean Square as coefficient update algorithm
A known adaptive algorithm such as the (LMS) algorithm is used.

Since the radiated sound of the muffling speaker 8 is also mixed into the detection signal of the sensor microphone 1, the muffling speaker 8
The sound field characteristic between the sensor microphone 1 and the sensor microphone 1 in advance, that is, the identification filter 13 is obtained by identifying the sound field characteristic.
The signal passed through the output signal of 1 is input to the minus side input terminal of the subtractor 4 and subtracted from the detection signal of the sensor microphone 1 to output the noise reduction speaker 8 to the noise reduction filter 5 input signal. Mixing, that is, howling caused by mixing is prevented. Further, the filter coefficient updating unit 12
The input data to the adaptive algorithm is the signal obtained by passing the output signal of the subtractor 4 through the identification filter 14 obtained by measuring or identifying the sound field characteristic between the muffling speaker 8 and the monitor microphone 9 in advance. By using, the speaker 8 for muffling and the monitor microphone 9 at the time of muffling
The sound field noise between is removed. The above is the same as the known technique except for the adder 17.

The features of the present invention are as follows. To the other input terminal of the adder 17 provided on the output side of the sound deadening filter 5, a minute signal that does not affect the sound deadening itself is input from the signal generator 16. As the type of signal, a band-limited random noise signal can be considered, but any other signal may be used as long as it has a flat frequency characteristic to some extent in the frequency range of the sound deadening target. The addition signal of the minute signal and the output signal of the muffling filter 5 is the filter 18 for identifying the sound field between the muffling speaker 8 and the sensor microphone 1 and the filter coefficient updating unit 1 of the identifying filter 18.
9 and the sound field identification filter 21 between the muffling speaker 8 and the monitor microphone 9 and the filter coefficient updating unit 22 of the identification filter 21.

At the same time, the minute signal is emitted as a sound from the muffling speaker 8 through the DA converter 6 and the amplifier 7 together with the output signal of the muffling filter 5, and the sensor microphone 1 and the monitor microphone 9 are used. To reach. Sensor microphone 1
The minute signal detected by is amplified by the amplifier 2, converted into a digital signal by the AD converter 3, and then input to the + side input terminal of the subtractor 20. The output signal of the sound field identification filter 18 between the muffling speaker 8 and the sensor microphone 1 is input to the-side input terminal of the subtractor 20, and the subtractor 2
The filter coefficient updating unit 19 updates the filter coefficient of the identification filter 18 so that the subtraction result at 0 always approaches the minimum value. As the coefficient updating algorithm, a known adaptive algorithm such as the LMS algorithm similar to the filter coefficient updating unit 12 of the muffling filter is used. The updated filter coefficient is sent to the identification filter 13 through the path a, and the filter coefficient of the identification filter 13 is sequentially updated. As can be seen from the above, the sound field between the muffling speaker 8 and the sensor microphone 1 can always be accurately identified during muffling, which is very effective in preventing howling.

On the other hand, the minute signal detected by the monitor microphone 9 is amplified by the amplifier 10 and converted into a digital signal by the AD converter 11, and then input to the + side input terminal of the subtractor 23. The output signal of the sound field identification filter 21 between the muffling speaker 8 and the monitor microphone 9 is input to the-side input terminal of the subtractor 23, and the filter coefficient is updated so that the subtraction result of the subtractor 23 always approaches the minimum value. Part 2
In 2, the filter coefficient of the identification filter 21 is updated. As the coefficient updating algorithm, a known adaptive algorithm such as the LMS algorithm similar to the filter coefficient updating unit 12 of the muffling filter and the filter coefficient updating unit 19 of the sound field identification filter between the muffling speaker 8 and the sensor microphone 1 is used. To use. The updated filter coefficient is sent to the identification filter 14 via the path b, and the filter coefficient of the identification filter 14 is sequentially updated. As can be seen from the above, the sound field between the muffling speaker 8 and the monitor microphone 9 can always be accurately identified during muffling.
It is possible to prevent the sound deadening performance from deteriorating with the lapse of time.

[0017]

【Example】

(Embodiment 1) In FIG. 2, the apparatus of the present invention, in a duct as shown in FIG. 3, uses random noise having a band limitation from 40 Hz to 500 Hz as noise, and after 5 minutes have elapsed from the start of muffling, At time t = 0 [sec], the state of the output signal of the (a) sound deadening filter 5 and (a) the detection signal of the monitor microphone 9 for 4 seconds after the temperature inside the duct is changed by 20 ° C. is shown. In FIG. 2, the output level of the muffling filter 5 is stable, and even if there is a change in the detection signal of the monitor microphone 9, that is, the sound pressure, such as in the vicinity of part A, approximately 0.55 seconds after the temperature change, the output level becomes steady immediately. Since it has returned to the state, the muffling is normally executed. In the case of FIG. 2, the muffling effect was about 13 dB.

Comparative Example 1 On the other hand, FIG. 4 shows a comparative example.
Using the same random noise as in Example 1, 5 minutes after silence was started by the conventional apparatus, time t = 0 [sec.
] Shows the states of (a) the output signal of the silencing filter 5 and (ii) the detection signal of the monitor microphone 9 for 4 seconds after the temperature inside the duct is changed by 20 ° C. In FIG. 4, due to the sudden increase in the output level of the sound deadening filter 5 caused by howling, the sound cannot be muted at the portion B about 3.2 seconds after the temperature change. Further, the detection signal of the monitor microphone 9, that is, the sound pressure, also increases considerably sharply in the vicinity of the portion B, and it can be seen that howling causes a large amount of noise.

As is clear from a comparison between FIGS. 2 and 4,
According to the device of the present invention, howling does not occur even if the sound field characteristic greatly changes after the start of muffling, and it is possible to prevent deterioration of the muffling performance. The space to which the noise reduction is applied may be a one-dimensional space such as a duct or a three-dimensional space such as a room or the outdoors.

[0020]

By using the device of the present invention, howling can be always prevented under any condition when active muffling is performed in a place where temperature change or wind speed change is severe, such as in a duct. Since the initial silencing performance can be maintained, it is expected that the reliability of the active silencing device will be greatly improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing an outline of an embodiment of the present invention.

FIG. 2 is a diagram showing a muffling situation by the device of the present invention.

FIG. 3 is an explanatory diagram showing a duct that has been muted.

FIG. 4 is a chart showing a muffling situation by a conventional device.

FIG. 5 is a block diagram showing an outline of an example of a conventional device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 sensor microphone 2 amplifier 3 analog-digital converter 4 subtractor 5 filter for noise reduction 6 digital-analog converter 7 amplifier 8 speaker for noise reduction 9 monitor microphone 10 amplifier 11 analog-digital converter 12 filter coefficient updating unit 13 identification filter 14 Identification Filter 15 Control Device 16 Signal Generation Unit 17 Adder 18 Identification Filter 19 Filter Coefficient Update Unit 20 Subtractor 21 Identification Filter 22 Filter Coefficient Update Unit 23 Subtractor

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Office reference number FI technical display location H03H 21/00 8842-5J

Claims (1)

[Claims] 1. A noise emitted from a speaker for noise reduction detected by a monitor microphone, the noise being reduced by creating a sound wave having the same amplitude and opposite phase as that of the noise detected by the sensor microphone by superimposing it on the noise. In the adaptive active noise suppressor that sequentially updates the characteristics of the noise reduction filter using the synthesized sound of noise and noise as an error signal, the characteristics of the noise field identification filter between the noise reduction speaker and sensor microphone used to prevent howling are sequentially updated. Means, a means for sequentially updating the characteristics of the sound field identification filter between the muffling speaker and the monitor microphone used for removing sound field noise, and a means for updating the two filter characteristic updating means for performing sound field identification. An adaptive active silencer, comprising: an input signal generating means.